Patients, Materials, and Methods

Patients and samplesAfter informed consent was obtained, the remaining aliquots of serum samples submitted for biochemical tests in patients visiting Chang Gung Medical Center were collected for this study. Samples from 4 groups of patients were included for NV-F sequence detection: (1) 180 healthy subjects (from Health Examination Service, Chang Gung Medical Center) with normal alanine aminotansferase (ALT) levels who were negative for HBV surface antigen (HBsAg), anti-HCV antibody, and HEV RNA; (2) 150 patients with hepatitis B who were positive for HBsAg and negative for IgM class anti-HAV antibody, anti-HDV antibody, anti-HCV antibody, and HEV RNA; (3) 150 patients with hepatitis C who were negative for HBsAg and IgM anti-HAV antibody, positive for anti-HCV antibody, and negative for HEV RNA; and (4) 69 patients with non–A–E hepatitis with serum ALT levels elevated >2.5-fold who were negative for HBsAg, IgM anti-HAV antibody, IgM class antibody against HBV core antigen (HBc), anti-HCV antibody, HEV RNA, and HCV RNA. None of these patients were alcoholics, and no known hepatotoxic medicine had been taken. Patients with fatty liver were not excluded from this study. All patients were negative for autoimmune markers, including anti-nuclear antigen, anti–smooth muscle antigen, and anti-mitochondrial antigen. In addition, patients were all negative for other virological markers, including antibody for HIV, IgM class antibody for Epstein-Barr virus, and IgM class antibody for cytomegalovirus. After the polymerase chain reaction (PCR) assays for the NV-F sequence, adequate amounts of samples were still available for the detection of anti–NV-F antibody in 155 patients. After informed consent was obtained, liver biopsy samples from 2 patients (patients F and B) whose serum was positive for the NV-F sequence were subjected to immunofluorescence analysis

Serological studiesHBsAg, IgM anti-HAV antibody, IgM anti-HBc antibody, and anti-HDV antibody were assayed using radioimmunoassay kits (Ausria-II, HAVAB-M, and anti-delta; Abbott Laboratories). Anti-HCV antibody was detected using an enzyme immunoassay kit (HCV-II; Abbott Laboratories). HCV RNA was detected by reverse transcription (RT) PCR assay (Amplicor HCV test; Roche Diagnostic Systems). HBV DNA was detected by Amplicor HBV Monitor Test (Roche Molecular Systems). The method of HEV RNA detection has been described elsewhere [3, 4]

Extraction of DNA or RNA, RT-PCR, and PCRTotal serum DNA was extracted using proteinase K digestion followed by phenol/chloroform extraction, as described in our previous publication [11]. Total serum RNA was extracted using TRI reagent (Molecular Research Center), in accordance with the protocol provided by the manufacturer. RT was performed using random primers. The procedure for RT and PCR has been described elsewhere [12]. Three primers were engineered: P1, 5′-CCGCGG(N)₄-3′; P2, 5′-GAATTC(N)₄-3′; and P3, 5′-GCTTGCTCTGTCTC(T)₂₀-3′. Each of the 4 Ns in P1 and P2 was a mixture of A, T, C, and G in equal ratios. After extraction of the total serum DNA or RNA from patient L, PCR or RT-PCR was performed, using random hexamers for 25 cycles; the product was then amplified using any 2 of the P1–3 primers. The resulting products were cloned into a vector, pCR2.1-TOPO (Invitrogen). For PCR detection of Escherichia coli 16S ribosomal DNA, the following primers were used: 16SL, 5′-GTCTGGGAAACTGCCTGATG-3′ (nt 121–140) and 16SR, 5′-GCTTCTTCTGCGGGTAACGT-3′ (nt 500–481)

Elimination of clones derived from the human genomeTo eliminate clones derived from the human genome, the clones were first lifted onto a nitrocellulose filter and hybridized with a mixture of probes generated from total liver RNA, as described in our previous publication [13]. Briefly, single-stranded probe was generated from cytoplasmic RNA extracted from normal human liver tissue. The tissue was minced into small pieces and lysed in a buffer containing 10 mmol/L Tris HCl (pH 7.2), 150 mmol/L NaCl, and 0.5% Nonidet P-40 (Sigma). After centrifugation at 1500 g for 5 min, the supernatant was used for RNA extraction. RT was performed using SuperScript II RNase H minus Reverse Transcriptase (Invitrogen), and oligo(dT) was used as the RT primer. One-third of the dTTP in the dNTP mixture was replaced by digoxigenin-11-dUTP (Boehringer Mannheim) to generate digoxigenin-labeled probes. The probes were mixed (molar ratio, 1:2) with oligo(dA) at 40°C for 1 h before hybridization. The hybridization signal was detected by use of a DIG Luminescent Detection Kit (Boehringer Mannheim). For each batch of hybridization, 1 ng of pCR2.1-TOPO without a cDNA insert was used as a negative control, and 1 pg of pCR2.1-TOPO containing a fragment of human albumin gene (Hs.184411) was used as a positive control. The negatively hybridized clones were considered to be of nonhuman origin

Automatic sequencingThe nonhuman-origin clones were subjected to automatic DNA sequencing (CEQ 2000; Beckman Instruments). The sequence data were further searched against the National Center for Biotechnology Information (NCBI) human genome data bank (http://www.ncbi.nlm.nih.gov/genome/seq/HsBlast.html), to eliminate human sequence

Development of anti–NV-F antibodyThe putative partial coding sequence of NV-F, flanked by NV-F1 and NV-F4 primers, was inserted into a vector, pYES2/NT (Invitrogen Corporation), and was arranged in-frame with the upstream polyhistidine region and the Xpress epitope sequence. The coding region of the whole fusion protein was subsequently isolated by restriction enzyme digestion (HindIII to XbaI), blunt-ended, and inserted into the SmaI site of pBacPAK8 (Clontech Laboratories). The fusion protein was expressed using the BacPak Baculovirus Expression System (Clontech). It was purified by a Ni²⁺-charged affinity column and was injected into a mouse for development of a polyclonal antibody. Alternatively, an initiation codon (ATG) was engineered in-frame with the putative coding sequence, and the resulting sequence was inserted into pBacPAK8, to express an NV-F peptide that did not contain any fusion parts. The primer used to generate the initiation codon (underlined) was 5′-ATGTGTTGGTGGCACAAAGCCC-3′

Immunofluorescence analysisFragments of liver specimens were snap frozen in isopentane cooled with liquid nitrogen and were stored at −70°C until use. Cryostat sections (5 μm) were dried at room temperature overnight and fixed in acetone at 0°C for 5 min. The immunofluorescence staining was performed using mouse polyclonal antibody against NV-F followed by fluorescein isothiocyanate–conjugated rabbit anti-mouse antibody (Jackson Immuno Research Laboratories). Double staining was performed by simultaneously staining the nuclei with DAPI (200 ng/mL). Confocal microscopy was performed using a Leica TCS SP2 Laser Scanning Spectral Confocal System

Results

Strategy to identify foreign sequences in the serum sample of a patient with non–A–E hepatitisA 66-year-old man (patient L) received a diagnosis of colon cancer (adenocarcinoma in transverse colon) in December 1999 at Chang Gung Medical Center. He received a colectomy, which was later complicated by anastomosis leakage, sepsis, and gastric ulcer bleeding. After intensive medical treatment, including blood transfusion, the patient’s condition was gradually stabilized. Unfortunately, an episode of acute hepatitis (peak ALT level, 284 U/L) with deep jaundice (bilirubin level, 19 mg/dL) occurred in July 2000. The patient was found to be negative for HBsAg, IgM anti-HAV antibody, IgM anti-HBc antibody, anti-HDV antibody, and anti-HCV antibody. The patient also tested negative for HEV RNA and HCV RNA. The serum sample obtained at this point was used for molecular cloning of foreign sequences

To identify foreign sequences in the serum sample, total serum DNA or RNA was extracted. The nucleic acid was then amplified (by PCR or RT-PCR) using random primers. The amplified product was subsequently subjected to a second-step PCR using designed primers (see Patients, Materials, and Methods). To eliminate sequence derived from human chromosomes, the resulting clones were hybridized with the probes generated from cytoplasmic RNA of normal liver tissue. All positively hybridized clones were discarded. The remaining 195 clones were sequenced using an automatic DNA sequencer. The sequencing data were compared with the human genome sequence, as well as with sequences in GenBank, by use of NCBI BLAST. Only 3 clones were found to be of nonhuman origin. One of the sequences, derived from the DNA extract, contained an open reading frame with incomplete 5′ and 3′ ends and was temporarily named NV-F (figure 1A). The sequence potentially encoded a peptide with incomplete amino- and carboxy-termini. Four primers, NV-F1 to NV-F4, were designed for the nested PCR assay. By use of this assay, this sequence was found to be absent in the chromosomal DNA extracted from HepG2 cells, Daudi cells, and 3 different sources of human peripheral blood mononuclear cells

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1

Identification of a foreign sequence in patients with non–A–E hepatitis. A Nucleotide sequence of the NV-F DNA fragment and conceptual translation of the putative partial reading frame. The positions of 4 primers (NV-F1 to NV-F4) used for polymerase chain reaction (PCR) detection of NV-F are marked with arrows. B Serum samples from patients with non–A–E hepatitis (lanes 1–9) patients infected with hepatitis C virus (lanes 10–19) patients infected with hepatitis B virus (lanes 20–29) and healthy individuals (lanes 30–36) subjected to an NV-F detection assay. Only part of the results is shown here. M, molecular weight marker; N1, negative control (NV-F–negative serum sample); N2, negative control (pure water). The arrow indicates the PCR product of NV-F

Detection of the NV-F sequence in patients with non–A–E hepatitisSerum samples from 4 groups of patients were included for the detection of the NV-F sequence (figure 1B). The sequence was detected in 5 (2.8%) of 180 healthy individuals. In contrast, NV-F was present in 17 (24.6%) of 69, 21 (14.0%) of 150, and 42 (28%) of 150 patients with non–A–E hepatitis, chronic hepatitis B, and chronic hepatitis C, respectively. One of the 17 patients whose serum was positive for NV-F had fulminant hepatitis. This was a 47-year-old male (patient F) who had non–A–E hepatitis accompanied by intermittent high fever and chills in May 2003. He was admitted for liver biopsy and further clinical investigation. Liver decompensation with massive ascites, bilateral pleural effusion, and consciousness disturbance developed 10 days after onset. Thereafter, the patient’s condition improved progressively without the need for any specific treatment, and he finally recovered completely. No known infectious agent was found throughout the course of the illness. Serial serum samples were obtained from this patient; his serum was found to be positive for the NV-F sequence during the early stage of the hepatitis flare, but it became negative thereafter (figure 2)

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 2

Clinical course in a patient with fulminant non–A–E hepatitis. Squares denote alanine aminotansferase (ALT) levels (U/L), and circles denote bilirubin levels (mg/dL). The periods of clinical symptoms are marked with solid bars, and “Bx” indicates the time of the liver biopsy. The NV-F sequence was detected by polymerase chain reaction in serial serum samples, and the results are indicated by a plus or a minus symbol

Expression of NV-F peptide and detection of anti–NV-F antibodyThe putative coding sequence flanked by NV-F1 and NV-F4 was used to express a fusion protein containing the putative NV-F antigen, polyhistidine, and an Xpress epitope, using insect cells. After purification, a doublet was found in the protein gel, which could also be seen by Western blot using anti-Xpress antibody (figure 3A). A mouse polyclonal antibody was then raised against the fusion protein. This antibody recognized a single protein species when only the NV-F peptide (no other fusion parts) was expressed in the insect cells (figure 3B, lane 1). By use of this peptide as an antigen, anti–NV-F antibody in serum from patient L was assayed. Western blot analysis revealed only 1 protein species (figure 3B, lane 3). The doublet derived from the fusion protein was, therefore, likely a result of partial degradation. Serum samples were subsequently examined for the presence of anti–NV-F antibody, using the insect cell lysate containing NV-F peptide (no other fusion parts) as well as the purified NV-F fusion protein as an antigen. The results obtained by use of the 2 methods were consistent. It was found that anti–NV-F antibody was present in 49 (75.4%) of the 65 patients whose serum was found to be positive for the NV-F sequence, including patient L and patient F. Of the 49 positive samples, 15 were from patients with non–A–E hepatitis, 16 were from those with chronic hepatitis B, and 18 were from those with chronic hepatitis C. In contrast, anti–NV-F antibody was undetectable in 90 patients whose serum was negative for the NV-F sequence (49 healthy individuals, 10 patients with non–A–E hepatitis, 11 patients with chronic hepatitis B, and 20 patients with chronic hepatitis C)

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 3

Generation of NV-F peptide and development of antibody against NV-F antigen. A A fusion protein containing polyhistidine, Xpress epitope, and a peptide encoded by NV-F was expressed in insect cells. The protein extract was purified by affinity column and was analyzed by electrophoresis. The molecular weight marker (M and M′), purified protein (lanes 1 and 1′), and nonpurified cell lysate (lanes 2 and 2′) were visualized by either coomassie blue staining (M, lanes 1 and 2) or Western blot analysis using anti-Xpress antibody (M′, lanes 1′ and 2′). The purified protein was then used to develop a mouse polyclonal antibody against NV-F. B The NV-F peptide alone (no fusion parts), subsequently expressed in insect cells. The cell lysate containing NV-F peptide (lane 1) and a mock control (lane 2) were analyzed by Western blot using the mouse anti–NV-F antibody. The cell lysate containing NV-F peptide was also analyzed, using a patient’s serum that was positive for the NV-F sequence (lane 3)

Characterization of the NV-F–associated agentThe nucleic acid was extracted from the serum sample from patient L, using either a DNA or an RNA extraction method. The nucleic acid was then digested by DNAse I, RNAse A, or S1 nuclease before the PCR assay. The results showed that the NV-F sequence was present only in the nucleic acid fraction that was extracted using the DNA extraction method. The NV-F sequence was resistant to RNAse A digestion but was sensitive to DNAse I and S1 nuclease digestion (figure 4A). To estimate the size of the NV-F–associated agent, the serum sample was mixed with 10⁵E. coli organisms and passed through a filter with a pore size of 0.2 μm. The nonfilterable material was resuspended in PBS and analyzed in parallel with the filtered portion. The result indicated that the putative particles containing the NV-F sequence were smaller than 0.2 μm (figure 4B)

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 4

Characterization of the NV-F agent. A Extraction of nucleic acid from the serum sample by either the RNA extraction (lane 1) or the DNA extraction method (lanes 2–4). After RNA extraction, reverse transcription (RT) polymerase chain reaction (PCR) for detection of the NV-F sequence was performed, without any intermediate step (lane 1). After DNA extraction, the extracted sample was treated with RNAse A (lane 2) DNAse I (lane 3) or S1 nuclease (lane 4) before subsequent PCR assay for NV-F sequence. M, molecular weight marker. The arrow indicates the PCR product of NV-F. B Size assessment of the NV-F agent. Serum containing the NV-F agent was mixed with Escherichia coli and passed through a filter with a pore size of 0.2 μm. PCR was performed to detect 16S ribosomal DNA of E. coli (lanes 1–4) or NV-F (lanes 5–8) in filtered (lanes 2, 4, 6 and 8) or nonfilterable (lanes 1, 3, 5 and 7) fractions. An aliquot of serum negative for the NV-F sequence (C1 and C2) was assayed in parallel as a mock control. The arrowhead indicates the PCR product of 16S ribosomal DNA, and the arrow indicates the PCR product of NV-F. C Cesium chloride gradient analysis for the NV-F agent. A serum sample positive for both hepatitis B virus (HBV) DNA and the NV-F sequence was used for cesium chloride gradient analysis. Twenty fractions were collected. All were sent for both HBV DNA quantitation (upper panel) and 1-step PCR (for the NV-F sequence) followed by Southern blot analysis (lower panel). Circles denote densities, and squares denote HBV DNA levels. The arrow indicates the PCR product of the NV-F sequence. N, negative hybridization control (1 ng of pCR2.1-TOPO); P, positive hybridization control (1 ng of the NV-F sequence)

It was found that, in some patients with chronic hepatitis B, coinfection with the NV-F agent and HBV occurred. A 36-year-old male (patient B from the chronic hepatitis B group) who had chronic hepatitis B with mild activity for >2 years came to our clinic to undergo a liver biopsy for fibrosis staging. A PCR assay revealed that his serum was also positive for NV-F. The serum sample from patient B was subjected to cesium chloride gradient analysis. The gradients were fractionated and assayed for the presence of HBV DNA (using a quantitative test) and the NV-F sequence (using 1-round PCR followed by Southern blot analysis). Two peaks of NV-F sequence were present, one in the fractions of 1.33–1.39 g/mL and the other in the fractions of 1.22–1.25 g/mL (figure 4C). The peak HBV DNA concentration was found in the fraction of 1.19–1.21 g/mL, indicating that the HBV particles were slightly lighter than the NV-F–associated particles. This experiment was repeated using serum samples from 3 other patients with NV-F–associated hepatitis, and the results were consistent

Immunofluorescence analysisBy use of the NV-F fusion protein expressed in insect cells, mouse anti–NV-F antibody was developed for immunofluorescence analysis. This antibody specifically detected the putative NV-F antigen (figure 3B). Immunofluorescence analysis was performed on the liver biopsy tissue obtained from patient F (figure 5) and patient B (figure 6). It was found that the antigen was distributed either in a speckle pattern or homogeneously in the cytoplasm of hepatocytes. Furthermore, positive staining was also observed in the perinuclear area (or on the nuclear membrane) in most positively stained cells

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 5

Detection of NV-F antigen in the liver biopsy sample from a patient with fulminant non–A–E hepatitis (patient F). Two different sections (A and B) from the same biopsy are shown. Positive cells in panel B are shown at higher magnification in panel C. Immunofluorescence analysis was performed using anti–NV-F antibody (left upper panel) and DAPI (right upper panel) for double staining. The pictures were overlapped using confocal microscopy (left lower panel). A negative control (right lower panel) using preimmune serum for staining was included. Scale bar, 20 μm

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 6

Detection of NV-F antigen in the liver biopsy sample from a patient with NV-F and hepatitis B virus coinfection (patient B). See the legend to figure 5 for further details

Discussion

Owing to technological advances in molecular biology, 5 major hepatitis viruses (HAV to HEV) have been discovered. The etiology of chronic hepatitis can thus be determined in a great majority of patients. Despite this achievement, the cause of chronic hepatitis remains elusive in ∼5% of patients [5, 14]. Furthermore, in acute hepatitis, the proportion of patients with undetermined etiology is even higher [3, 15]. In Taiwan, the HBV carrier rate is ∼15%, and more than half of the inpatients in Taiwan with acute hepatitis are seropositive for HBsAg [3]. It is believed that acute exacerbation of hepatitis B in chronic HBV carriers is responsible for the majority of acute hepatitis flares [16]. Even though the proportion of patients with non-B hepatitis is small, the etiology of acute hepatitis remains undetermined in 15.9% of our inpatients, suggesting the existence of other, unidentified hepatitis viruses [3]. In this study, we have identified a fragment of DNA sequence (NV-F) in the serum of a patient with non–A–E hepatitis. Only 2.8% of healthy individuals carried this sequence in their serum, whereas 24.6% of patients with non–A–E hepatitis were positive for NV-F. In this study, we did not exclude patients with nonalcoholic steatohepatitis from the non–A–E hepatitis group, nor did we exclude patients with fatty liver [17, 18]. It is possible that the prevalence of NV-F would be even higher if such patients were excluded. Interestingly, a high prevalence of NV-F is also observed in patients with chronic hepatitis B or C, indicating that coinfection with NV-F and either HBV or HCV frequently occurs. Similarly, when HCV was initially discovered, many studies on the seroprevalence of HCV indicated that HCV was found in >10% of HBV-infected patients worldwide [19]. The prevalence might be underestimated, since HCV superinfection exerts a suppressive effect on HBV and enhanced seroclearance of HBV [20]. Coinfection with HBV or HCV was also commonly found in patients with GBV-C, TTV, and SEN virus infection. Supposedly, such a high percentage of coinfection is attributed to a common transmission route. The effect of NV-F superinfection on chronic hepatitis B or C is not clear at this time. A detailed clinical analysis is needed to answer this question. Despite a high prevalence of the NV-F agent in non–A–E hepatitis, it is still questionable whether NV-F is the direct cause of hepatitis. Since NV-F frequently coinfects with HBV or HCV, it remains possible that NV-F coinfects with a yet-unidentified virus in patients with non–A–E hepatitis and that it is the unidentified virus that serves as the direct cause of hepatitis. In this study, we have provided 2 pieces of evidence suggesting that NV-F might contribute, at least in part, to the hepatitis activity. First, in patient F, NV-F viremia occurred concurrently with the hepatitis flare, and the NV-F agent was cleared from the serum after recovery from the disease. This temporal relationship argues for a causative role of NV-F in non–A–E hepatitis. Second, the NV-F antigen was found in the cytoplasm of hepatocytes, suggesting that this agent is hepatotropic. The presence of a foreign antigen in the liver cells frequently results in an inflammatory reaction—namely, hepatitis—unless other unknown mechanisms are involved to deter the host immune response. Further immunological study is needed to understand the mechanism of NV-F–associated non–A–E hepatitis

At this time, the biological nature of the NV-F agent has not been completely defined. Our data indicate that it is smaller than 0.2 μm, forms 2 buoyant densities in a cesium chloride gradient, and possesses single-stranded DNA. These features suggest that the NV-F agent is possibly a virus. The presence of 2 densities in cesium chloride gradient analysis is sometimes observed in an enveloped virus. A possible explanation is that some particles containing only the nucleocapsid (but not the envelope) form the band with the higher density. However, owing to an extremely low serum concentration of NV-F, the attempt to visualize the particles by electron microscopy failed. Southern and Western blot analysis using the remaining liver biopsy samples submitted for this study from patients L and B (only 3 mm in length) failed to demonstrate the viral genome and protein. A larger piece of tissue, such as surgically removed liver tissue, may be required to achieve this goal. A BLAST search showed that none of the known sequences shared sequence homology with NV-F. Further extension of the 5′ and 3′ ends of the NV-F sequence is, thus, progressing very slowly. The best-known single-stranded DNA viruses are parvoviruses and circoviruses. It is possible that NV-F belongs to a class of virus distantly related to one of these 2 families. Alternatively, it may represent a new class of agents that has no known close relatives

In summary, we have discovered a novel single-stranded DNA sequence that is associated with human hepatitis, including in a patient with fulminant non–A–E hepatitis. The NV-F agent is hepatotropic and likely belongs to a novel class of viruses. Finally, this virus frequently coinfects with HBV or HCV in patients with chronic hepatitis